[Corpora-List] Can corpora help to distinguish a dialect and a language?

[CRuehlemann at aol.com]

It is interesting to see how a serious question sparks a funny discussion,  
which is somewhat carnivalesque even, featuring soldiers, rabbis, schmucks, 
 dead dogs and what have you galore. On a more sober note one might return  
to the original question (Can corpora help distinguish dialect and a 
language?)  and say yes, to an extent, provided they have been carefully marked up 
 for  social categories. One such well-designed and annotated corpus is the 
BNC. In  pursuit of the above question, it can be used in two basic ways. 
In  a top down approach, one may start with a set of pre-defined dialect  
features, run queries for them using relevant restrictions such as speaker  
location or speaker dialect. Such a query would show, for example, that the  
quotative phrase *I says* is found predominatly in northern dialect areas of  
Great Britain (and Ireland). A bottom up approach would be to start without 
a  set of pre-defined dialect features and instead examine a set of randomly 
 selected features using the same set of restrictions as above. Such a 
query type  might, perhaps, help discover dialect features that researchers have 
not  yet been aware of as dialect features.
 
Apologies for this totally unMardi-Graslike note.  
 
Cheers
 
Chris
 
--------------------------------------------------------
Dr. Chrstoph Rühlemann, Munich
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listserv.linguistlist.org/pipermail/corpora/attachments/20100216/77454b21/attachment.htm>
-------------- next part --------------
_______________________________________________
Corpora mailing list
Corpora at uib.no
http://mailman.uib.no/listinfo/corpora